Update regarding Vercel service disruption on October 20, 2025
By Steven Van ·
Vercel says a cascading failure during AWS's us-east-1 outage briefly disrupted traffic serving and later knocked its dashboard and APIs offline for hours.
Vercel published a detailed account of the October 20, 2025 outage that followed AWS's us-east-1 disruption, saying it is "unequivocally responsible" for the incident regardless of what triggered it. The company describes the event as two independent incidents on its side.
- The initial loss of the us-east-1 region disrupted serving of production traffic. Vercel was alerted at 06:55 UTC and began rerouting traffic at 07:15 UTC, but a cascading failure at 07:45 UTC took down part of its global caching infrastructure, causing static file serving failures that peaked at about 22% of traffic. Service was restored by 08:18 UTC, and function invocations for customers relying solely on us-east-1 came back at 09:21 UTC.
- A separate outage of Vercel's feature flag provider at 19:20 UTC, described as a downstream effect of the AWS disruption, exhausted resources in Vercel's primary Kubernetes cluster and disrupted its control plane, including the dashboard, CLI, API and log forwarding. This left customers unable to deploy or roll back, though it did not affect production traffic. Vercel began rolling out a mitigation at 19:48 UTC and reports full restoration occurred around 03:00 UTC.
The Vercel post also walks through how a typical request is handled across DNS, CDN, TLS termination and firewall layers to show where the failure occurred. Full details are in the original announcement.
Vercel
The platform for frontend developers — deploy, preview, and scale web apps and AI agents with zero config.
View Vercel →